The FIFA rankings are terrible.
Behind the disguise of weighting factors and other funny math, the FIFA rankings are a table of results. Win a match, get 3 points. Draw a match, earn 1 point. Attempt to adjust for strength of competition and divide by the number of games played, and you have a rating for each country.
But tables are misleading. For example, Newcastle finished 5th in the English Premier League. However, they scored only 5 more goals than they allowed, only 8th best among 20 teams. Goal differential is predictive of team strength, so Newcastle got a bit lucky this year.
At least in the Premier league, each team plays every other team twice. Tables are even more misleading in world football since countries play schedules of differing strength.
Here at The Power Rank, we have developed better ranking system based on years of research in statistical physics. Instead of counting up points for wins and ties, we solve a set of linear equations, the bedrock operation in modern data mining. Moreover, our algorithm accounts for margin of victory and strength of schedule in ranking countries. Using all major international competitions since 2009, our algorithm assigns each team a rating, which gives a team’s strength in goals compared to the average international team. So Spain’s 2.0 rating says they will beat the average team (Belgium) by 2 goals on average at a neutral site.
While we haven’t tested these predictions on international football yet, the predictions for American college football work quite well. Over the last 10 years of college football bowl games (late season games played at neutral sites), the rankings have predicted more game winners, 62.4%, than the Vegas betting line, 61.7%. This sample includes 314 games.
For world soccer, the rankings reveal some hidden truths that everyone ought to know.
Brazil is the best team in the world
They haven’t done so well in their last two major competitions. The Brazilians exited the World Cup in 2010 in the quarterfinals against the Netherlands. Coach Dunga left after this debacle. Then in last year’s Copa America, Brazil lost to Paraguay in the quarterfinals.
However, our algorithm still thinks highly of Brazil due to their consistent track record of success. They won the Confederation’s Cup in 2009. More importantly, they finished first in World Cup qualifying out of South America, a brutal stretch of 18 games against some of the best teams in the world. Lastly, tournament soccer leaves even the best teams in the world subject to random chance. Brazil lost to the Netherlands on a freak own goal by Felipe Melo. Their exit in Copa America came in penalty kicks after a 0-0 draw.
Let’s contrast Brazil with Uruguay, the 5th best team in the world. Uruguay had more success than Brazil in the World Cup (4th place) and Copa America (winner). However, they finished 5th in World Cup qualifying in South America, 10 points back of Brazil. Uruguay had to win a playoff against Costa Rica just to make the World Cup.
And then no one will ever forgot the luck bestowed upon Uruguay against Ghana in the World Cup. Tied late in the quarterfinal game, Uruguayian striker Luis Suarez intentionally blocks a sure goal with his hand. Then Ghana’s Asamoah Gyan missed the ensuing penalty kick, allowing Uruguay to win the game in penalty kicks. Watch it again here.
We mean no disrespect to Uruguay. It takes incredible talent and work to earn the 5th spot in these rankings. However, Brazil is the better team.
Is the United States better than Mexico?
Well, no. The Power Rank puts Mexico at 12th, while the United States comes in at 22nd.
However, this distinction rests on a single game. In 2009, the United States went to the Confederations Cup in South Africa. That summer also featured the Gold Cup, the North American championship held every two years. While the Americans sent their first team to South Africa, they played a reserve unit on home soil in the Gold Cup. The reserves played very well, making it all the way to the final against Mexico. Even through halftime, the championship remained tied at 0.
Then the wheels came off for the Americans. Mexico, who played first teamers like Gio Dos Santos and Gerardo Terrado, scored 5 goals in the second half. The U.S. line up featured Jay Heaps and Brian Ching, players who would not make the 2010 World Cup squad.
If we do not include games from the 2009 Gold Cup, the United States and Mexico are ranked 18th and 19th respectively. Clearly, better rankings would account for situations in which one team doesn’t play their first team. For now, we’re treating all international matches the same.
Who is the best team in Europe?
Spain and the Netherlands are 2nd and 3rd in the rankings, separated by less than a hundredth of a goal. Essentially, our algorithm does not make a distinction between these two teams. The finals of Euro 2012 could quite possibly be a rematch of the World Cup final two years ago.
Next week, we’ll reveal the full rankings and discuss how to use our algorithm to project the entire bracket for Euro 2012. If you’re interested in this, please follow The Power Rank on Twitter.
What do you think?
We would love to hear from you. Are there other games like the 2009 Gold Cup final in which one team didn’t play their first team? How necessary is it to include friendly matches in these rankings? Please leave us a comment.
Thanks for reading.
Friendliest should not be included or you should have a duplicate set of stats showing rankings without friendliest.
Thanks for reading, Anthony. The main argument for including friendlies is that they provide more matches between continents. When I first did these rankings right after the World Cup group stage in 2010, I had only those group stage games and the Confederations Cup to “mix up” the world. With the South American dominance in the group stage, I think Venezuela was like 12th. Clearly, this fixed itself after the knock out stage.
In these rankings, I have a larger set of games, so I feel relatively comfortable with how the European and South American teams are ranked. But I will take you up on your suggestion and offer rankings with and without friendlies.
I like the new font, Ed.
Thanks for reading, Ted. The font is a product of Genesis, the new WordPress framework I’m using. I bought it for security and search engine optimization, but I’ve been most happy with the choice of fonts. The basic design made it very easy for me to cook up this design.
I think the concept is good and looking at the rankings I would not have much argument with them. But my thinking behind friendlies being excluded is that managers use friendlies to try new players and often give uncapped players a run out. In some of these games the managers may also make eleven or more changes. To me friendly results would distort the rankings somewhat. I do see why you included them given the need for crossover but in essence those crossover games often involved trying out new players and trying new formations. Great work though on the rankings.
All good points. Eventually, I’ll test a few options with friendlies and post the rankings with the most predictive power.
Hey Ed, nice comprehensive post. Friendlies are definitely an issue. Leave them out and you lose crucial information. Weight them too heavily, and the rankings are sensitive to the result as there are so many more friendlies. Egypt is the most intriguing team at #13 and have definitely had some crazy luck against them. I’m surprised Denmark is so high as well. The funniest result has got to be France not even cracking the top 25. For their talent, they must be one of the must under-producing international teams in history. Speaking of luck, though, a rough estimate of Chelsea surviving both Messi and Robben penalty kicks is about 1%. If Di Matteo keeps his job, I’m shorting Russian oil.
A better way to account for Friendlies as well as include the vast mountain of data from professional league play might be to solve a system of equations based on players instead of teams.
For example, since 2009 Leo Messi has played in 168 games for Barcelona in all competitions, but only 34 for Argentina. Increasing the sample of data by five fold would surely improve the predictive power of the data, but the complexity of including almost every player on the planet might not be feasible nor worth the time and effort.
Thanks for reading. There are some problems with solving a system of equations for players. The first one is that not all of them have a clear metric, goals, that correlates to winning. However, there are some systems that do include player data, like SPI at ESPN.
http://soccernet.espn.go.com/spi/rankings
I’m not a huge fan of these ranking, since they tend to overvalue countries like England with talented players who don’t get it done on the international stage. Moreover, they have Paraguay grossly underrated.
Great content material and great layout. Your blog deserves all the positive feedback it has been getting.
Hey Ed,
Great post.
As far as friendlies are concerned, they should not be factored in. By definition they are merely exhibitions, in other words devoid of competitive value.
Just have a few questions,comments concerning the Mexico/US ranking comparisons. Did you exclude Mexico’s 2011 Copa America results? Per a CONCACAF mandate Mexico was only allowed to field its U-23 team. Whereas for the 2009 Gold Cup, no such restriction was placed on the US. The US simply chose not to field its first team. Therefore the 2009 Gold Cup final should be treated just as the 2011 final.
From January 2009 to the present, Mexico & the US each participated in five official competitions. Mexico: 2009 GC, 2010 WCQ, 2010 WC, 2011 GC & 2011 Copa America. US: 2009 Confed Cup, 2009 GC, 2010 WCQ, 2010 WC & 2011 GC. Therefore to get the best oranges to oranges comparison, I proposed that you exclude Mexico’s 2011 Copa America & the US 2009 Confederation Cup results from your analysis. Your thoughts?
Thanks for reading, Adrian. In short, I did include Mexico’s 2011 Copa America, and I agree that Mexico and US games from 2011 Copa and 2009 Gold Cup respectively should be excluded. (Think you had a typo above.)
I do plan to implement all of this. Just a matter of getting the code together. I think Costa Rica sent their U23 squad to 2011 Copa America.
Ed,
That was not a typo. When strictly comparing only Mexico/US, I propose that you exclude the 2011 Copa America for Mexico & the 2009 Confederations Cup for the US.
The 2009 GC for the US should be included, simply because the US chose not to send it A squad.
Ok, then I don’t agree. I think we should include every game in which a country plays its best team. So I understand why you want to exclude Mexico in Copa America 2011. But the US played their first team in Confederation’s Cup 2009. Why not include it?
Of course, this is my American bias coming out a bit. The US played well in Confed Cup 2009, and excluding these games would certainly drop their ranking.
Ed,
My point it to get an apples to apples comparison for Mexico/US only. Mexico participated in the 2011 Copa America, the US didn’t. The US played in the 2009 Confed. Cup, Mexico didn’t. So based solely on results from the other 4 competitions in which both participated, how wide is the gap?
Now to compare the US to the rest of the world, of course include the Confed Cup.
Got it, thanks!
I love ranking systems.
As a mathematician, and a huge football fan, let me just congratulate you first on having developed the Power Rank. This is a dream endeavour and I appreciate the effort and all the goodies you are providing us here.
Some thoughts on the FIFA ranking.
I would like to point out some positive things we can take from the current iteration of the FIFA Rankings. They take the last 4 years of games and diminish the importance of the game the older it gets. The strength of the opponent is considered in the calculation, therefore a win versus Spain, is better than a win versus Bolivia. And also, the importance of the game is a significant factor that contributes to points. Friendlies are counted, but not as important as major competitions.
On the other hand, sadly, the score of the game has been dropped from this version of the FIFA rankings. Older versions included a score factor but it might have been deemed “too complex” and has been discarded. Home and away is not considered as a factor, although, again, I believe there was a consideration for this earlier versions. I find that these two components would add to the ‘accuracy’ of the rankings.
There is some good, and there is some bad. Clearly Ed, your ranking and the FIFA ranking match up the top 5 teams in the world and that in itself should be a sign that some things have been done correctly by both. I would agree with you that FIFA has missed some opportunities to enrich the ranking system (as stated above) but I would like to point out some interesting facts.
Paraguay
Paraguay, ranked 8th here and 22nd by FIFA are on a terrible run. They have lost to Peru, Bolivia, Chile, and managed a home tie (scoring in the 93rd minute to salvage the draw) all in the past year. Losing by two goals, against Bolivia and Peru in WC qualifying is a huge disappointment. I feel that FIFA does penalize Paraguay for these results and they are no longer a top 20 team (according to FIFA) at this moment.
Chile on the other hand have been in good form having won 4 straight in WC qualifying and I feel that their top 10 ranking here, and on FIFA, is justified. I would say that Chile is a better team than Paraguay right now. Not many people in South America would disagree. Chile tops the WC qualifying group (albeit with a game at hand) and Paraguay is near the bottom of the table. It is early days (as you mentioned it is a gruelling 2 year process) but my conclusion stands. Chile is currently better than Paraguay.
On Uruguay:
I would like to acknowledge your “no disrespect to Uruguay” statement. I appreciate this sentiment. I would like to point out that mentioning the Ghana game does seem to ‘devalue’ Uruguay’s worth in your Uruguay vs Brazil comparison. If I may…
To call the hand-ball-goal-line save by Luis Suarez luck, is both unfair, and a complete understatement. Let me explain.
If we rewind a few seconds before Suarez’s handball, a Ghanaian had a chance inside the box with no goalie. It was saved by Suarez with his feet. If we rewind a bit more you would see the players were in an offside position in more than one occasion. You can see it here:
http://www.metacafe.com/watch/4859291/gyan_offside_uruguay_ghana_120min_not_penalty/
Also, if we rewind a bit more, you would see that the free kick that induced the whole play was not really a foul in the first place. It was a dive from the Ghanaian in an effort to get a call because he had no way of getting the ball. I’m not claiming a conspiracy, but a lot of luck (hence the understatement) had to go the way of Ghana before the handball ever happened. Yes, Uruguay might have been considered lucky with the handball incident, but Ghana did not take advantage of all their own luck.
And lastly, (because I want to get back to the math) Uruguay was badly damaged by the Ghanaians, losing their captain, two starting defencemen, and Luis Suarez for the next match versus Netherlands. What would have happened if there was a fit Uruguay team against Netherlands, it does not matter. But to ‘devalue’ Uruguay because of the Ghana game, is unfair.
Back to the rankings.
I do believe that Uruguay should be granted a top 5 spot because of their performance in the last 3 years. Your ranking here has done that. They are currently 2nd in the world on FIFA but will quickly drop to 4 or 5 after the Euro. That too, is a feature of the FIFA rankings that I commend. You might disagree on this point. South American teams would have been on the rise because of last year’s Copa America, but now the European teams (Germany, England, Portugal) will bounce back up because of their very recent (present day in the case of Portugal) results.
A quick look at Portugal v Denmark.
Portugal and Denmark were in the same group during Euro qualifying and Denmark’s results were better than Portugal’s. But, it was close. Denmark are currently ranked higher than Portugal on FIFA, but it’s close. Portugal’s results in Euro will surely propel them above Denmark in the next FIFA ranking. Again, I enjoy this aspect of the FIFA ranking.
Of course, as always, none of this can be looked at in a vacuum and the more data points you have the richer the results. This is where your Ranking has the advantage.
In conclusion, I would like to theorize that what the FIFA ranking does is reward teams for results both current and in the long run (4 years) which is not exactly the same as determining which is the best team. Was Greece the best team in 2004? I would say not. But their results were rewarded, both in being crowned as Euro champions and in a 20 or so rank jump on FIFA. I commend your efforts here with the Power Rank. Your goal of trying to determine the better team is mathematically an intriguing endeavour. I also respect the goal of attempting to predict the results. You are, in a literal sense, looking at the past to predict the future. Whereas FIFA is looking at the past, and rewarding the results.
Congratulations again on the great work. I look forward to following you and the Power Rank over the next few months to see where it will take us.
Nico
No way is Brazil better than Spain. Spain are clearly the best team in the world, and have been so for 4 years.
Thanks for reading, fellas.
With their performance against Italy in the Euro 2012 Final, Spain is now the top team in the rankings.
https://thepowerrank.com/world-football-soccer/
Spain’s defense is amazing and young, which means we should all be bullish on them heading towards Brazil 2014.
This post is in fact a nice one it assists new net visitors,
who are wishing for blogging.